ACDEFGHIJKLMNOPQRSVWXYZAA
1
Model Name#Parameters
(Millions)
Model Size
(MB)
Model
Performance/Size
Ratio
GLUE Datasets
2
GLUE
Score
CoLASST-2MRPCSTS-BQNLIQQPMNLI-mMNLI-mmRTEWNLI
3
Matthew'sAccuracyF1AccuracyPearsonSpearmanAccuracyF1AccuracyAccuracyAccuracyAccuracyAccuracy
4
BERT Models
5
BERT-base109389.710.70677.052.193.588.985.890.571.284.683.466.453.5
6
BERT-large3341187.840.24180.560.594.989.385.487.686.592.772.189.386.785.970.165.1
7
DistilBERT66235.971.11573.645.892.387.683.1717188.869.688.281.681.354.165.1
8
TinyBERT66235.971.18378.151.193.187.382.68583.790.471.689.184.683.27065.1
9
10
RoBERTa Models
11
RoBERTa-base1250.69186.363.694.890.291.292.891.887.678.7
12
RoBERTa-large355796.140.24888.167.896.792.389.892.291.995.474.390.290.890.288.289
13
14
ALBERT Models
15
ALBERT-base1242.66.98683.865.192.988.086.087.286.789.769.985.384.684.186.5
16
ALBERT-large1864.154.76085.766.694.990.087.889.188.691.771.587.286.585.988.4
17
ALBERT-xxlarge235789.890.37989.269.197.193.491.292.59295.274.290.591.39189.291.8
18
19
ELECTRA Models
20
ELECTRA-small1448.295.70779.964.186.881.182.785.681.281.679.882.7
21
ELECTRA-base110385.760.77485.168.392.486.388.191.286.486.985.088.1
22
ELECTRA-large3341187.840.26889.471.797.190.792.595.890.891.389.392.5
23
24
MiniLM Models
25
MiniLM-L6xH3842246.43.64480.247.591.588.990.590.683.368.8
26
MiniLM-L12xH3843367.72.52383.358.59389.591.591.385.773.3
27
28
T5 Models
29
T5-small602311.29077.44191.889.786.685.68590.3708882.482.369.969.2
30
T5-base220779.420.37682.751.195.290.787.589.488.693.772.689.487.186.280.178.8
31
T5-large7702764.80.11286.461.296.392.489.989.989.294.873.989.989.989.687.285.6
32
T5-3B2800245760.03288.567.197.492.59090.689.896.374.489.791.491.291.189.7
33
T5-11B11000409600.00890.371.697.592.890.493.192.896.975.190.692.291.992.894.5
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100